validate various displayio args; new pin validation routines; don't use mp_const_none if NULL will do

[dhalbert]

There was some missing pin argument validation in displayio, and when starting to fix that, I decided to simplify how pin argument validation and free pin checking was done.

• Replace assert_pin completely, and subsume most uses of assert_free_pin() by adding the new routines validate_is_free_pin() and validate_is_free_pin_or_none(), which check whether an mp_obj_t is a pin and return an mcu_pin_obj_t * if it is. This reduces several lines of code to one, improving readability and reducing boilerplate.
• Be consistent that optional *_obj_t arguments passed to common_hal_... (mostly pins) are passed as NULL instead of mp_const_none. Also, use of mp_const_none in common-hal is minimized.

The changes save about 260 bytes in the Trinket M0 build.

[tannewt]

I didn't look at the code (add me as a reviewer or @ me when ready) but I think validate_pin_is_free would read better.

[dhalbert]

I didn't look at the code (add me as a reviewer or @ me when ready) but I think validate_pin_is_free would read better.

The old assert_pin_free() did not actually check that the object was a pin, so I was trying to make that clearer. But I agree the name was a bit awkward. I think maybe validate_obj_is_pin() and validate_obj_is_free_pin() would be more grammatical and clearer.

[tannewt]

validate_obj_is_free_pin works for me. Thanks!

[dhalbert]

The churn of trying checkout@v2 (see #2683) and then reverting it fixed the build problems. (EDIT: not really)

[dhalbert]

Now I am getting a "ERROR: THESE PACKAGES DO NOT MATCH THE HASHES FROM THE REQUIREMENTS FILE." errors on a few random builds, perhaps because of VM problems??

[tannewt]

One minor improvement. Good otherwise. Is the CI simply busted now?

[tannewt]

Thank you!

[tannewt]

FYI this breaks Trinket M0 and likely any other boards that use a DotStar as the status LED. The SPI constructor call needs to be updated to pass NULL instead of mp_const_none here: https://github.com/adafruit/circuitpython/blob/master/supervisor/shared/rgb_led_status.c#L128

[dhalbert]

I looked through all the mp_const_none uses, but clearly not for all values of "all". :)